a b s t r a c t

Objectives: The primary objective of this study was to determine the proportion of patients with medica- tion discrepancies when using a self-administered medication history form in the emergency department (ED). The secondary objectives were to identify predictors of medication discrepancies and determine the proportion of patients with a high-risk medication discrepancy.

Methods: This was a cross-sectional study conducted in an urban ED in Australia. Patients completed a self-administered medication history form while waiting to be seen by a physician. Subsequently, a best possible medication history was taken by a pharmacist to determine accuracy of the self-reported med- ication lists for patients with planned admissions. Discrepancies between the two medication lists were reported descriptively. A Poisson regression analysis was conducted to identify predictors of the rate of discrepancies. Associations were reported as incident rate ratios (IRR).

Results: A total of 138 patients were included in the study. The total number of discrepancies was as follows: 0 (25%, n = 34), 1 (34%, n = 47), 2 (11%, n = 15), and >=3 (30%, n = 42). The number of medications

(IRR 1.11, 95% CI 1.09 to 1.14, p < 0.001), female (IRR 1.51, 95% CI 1.18 to 1.92, p = 0.001), and missing

community pharmacy information (IRR 2.10, 95% CI 1.64 to 2.68, p < 0.001) were significantly associated with rate of discrepancies. Overall, 20% (n = 28) of patients had one or more high-risk medication discrepancies.

Conclusion: Patient self-administered medication history forms have a high rate of discrepancies and should be verified by a best possible medication history.

(C) 2019

1. Background

A medication history must be taken upon patient’s arrival to the emergency department (ED) for initial Clinical decisions and ongo- ing care [1]. The initial history affects the reconciliation process, influencing the appropriateness of medications that should be con- tinued upon hospital admission and subsequent discharge [2]. However, the process of history-taking is time consuming, espe- cially when patients have complicated histories. In one time-and- motion study, the process involved up to three professionals, from

* Corresponding author at: Pharmacy and Bank Building (A15), Camperdown Campus, University of Sydney, Sydney, New South Wales 2006, Australia.

E-mail addresses: Angela.Wai@health.nsw.gov.au (A. Wai), Martina.Salib@ health.nsw.gov.au (M. Salib), Sohileh.Aran@health.nsw.gov.au (S. Aran), James. Edwards@health.nsw.gov.au (J. Edwards), asad.patanwala@sydney.edu.au (A.E. Patanwala).

two different specialties, and took >90 min for some patients [3]. Thus, there is great interest in the ED to improve efficiency in this process. One possible solution has been for patients to complete a self-administered medication history form while they wait to be seen by a physician [4]. Although this practice is widely accepted in the Primary care setting, it is not common in the ED. In one study in the U.S. up to 70% of patients who presented to the ED were able to complete a self-administered form, suggesting that this approach can be adopted in this setting [5]. However, there is limited data regarding the accuracy of using a self- administered form in the ED.

In one study involving 160 ED patients in Canada, most patients were able to complete a self-administered medication history form with no clinically significant discrepancies [6]. However, the sam- ple included only non-admitted patients. Therefore, it is difficult to extrapolate these results to those who are more acutely ill and who

https://doi.org/10.1016/j.ajem.2019.04.016

0735-6757/(C) 2019

require admission. The non-admitted patients may closely resem- ble those who are seen in General practice, for whom there is less controversy for a self-administered form. In Australia, the Clinical Excellence Commission Continuity of medication management Expert Advisory group recommends that a medication history should be documented within 24 h of hospital admission, high- lighting the need for data in this subset of patients [7]. This data is necessary as self-administered forms are implemented in the ED in the future.

The primary objective of this study is to determine the propor- tion of patients with medication discrepancies when using a self- administered medication history form in the ED. The secondary objectives were to identify predictors of medication discrepancies and determine the proportion of patients with a high-risk medica- tion discrepancy.

1. Patients and methods
   1. Study design and setting

This was a cross-sectional study conducted in an urban ED in Australia. The ED has close to 80,000 annual patient visits and patients who present to the ED have a diverse socioeconomic sta- tus. The Ethics Committee of the institution approved the study prior to implementation (Ethics approval number X17-0106 & LNR/17/RPAH/152). A quality improvement process was imple- mented in the ED, whereby patients and/or their caregivers were provided with a self-reporting medication history form called ‘MYMEDS’ (Appendix) to improve the medication history-taking process. Patients completed the form while waiting to be seen by a physician. Subsequently, a best possible medication history (BPMH) was then taken by a pharmacist to determine accuracy of the self-reported medication lists provided by patients. Only patients with a BPMH were included. A BPMH is defined as system- atic process of interviewing a patient/carer to obtain a medication history. The process involves utilizing at least at least one other reliable source of information (e.g. general practitioner or commu- nity pharmacy records) to obtain an accurate list [8]. The pharma- cist was available to complete the BPMH from 0900 to 1630 h from Monday to Friday.

Patient selection

Patients who presented to the ED were selected during a 2- week study period. Upon initial presentation to the ED, a nurse or pharmacist approached patients for consent. Patients who were in the resuscitation area were not approached (i.e. Australian triage category 1). Inclusion criteria were age >= 16 years, able to commu- nicate in English or accompanied by a caregiver who could com- municate in English, and with a plan for hospital admission. Patients were excluded if they had psychiatric or behavioral symp- toms as determined by the investigators, had impairED capacity (e.g. critically ill) or were from an aged care facility. Patients from aged care facilities are transported with medications lists from these facilities. Thus, the accuracy of self-reporting is less applica- ble in these circumstances.

Data collection and variables

Data was extracted from the MYMEDS form and the BPMH for each patient and entered into a web-based form (Research Elec- tronic Data Capture) [9]. Data collected pertained to demographics

(age, sex), whether or not patients completed certain fields on the form (e.g. allergies, community pharmacy contact details, general practitioner contact details), and medication information. Two investigators reviewed the MYMEDS forms and BPMH to quantify discrepancies. Any disagreement was resolved via discussion. From an ED perspective, discrepancies were determined at two levels. First, whether or not the medication names provided were accu- rate. Accuracy was determined based on whether the wrong drug was listed or omitted. Spelling mistakes were not counted as errors. Second, if the medication name was accurate then whether or not an accurate regimen was listed. The regimen was defined as a combination of the strength (e.g. 10 mg), amount (e.g. number of tablets), and frequency (e.g. every morning) so that a complete pre- scription could be ascertained. Some medications have only one strength, amount, or frequency that is possible for any labelled or off-labelled indication. A missing value in these circumstances could still be considered to be a complete regimen based on assess- ment by the investigators. The total number of discrepancies was calculated based on drug names and regimens. If a drug name was omitted it was considered to be one discrepancy because the omission of the regimen is assumed. Any error in a regimen (i.e. dose, frequency) that would result in the wrong amount provided to patients was considered one discrepancy. No more than one dis- crepancy could be counted per regimen. Discrepancies were cate- gorized as high-risk based on the definitions by the Clinical Excellence Commission [10]. A high-risk medication is defined based on the pneumonic ‘A PINCH’: anti-infectives [A], potassium and electrolytes [P], Insulin or antidiabetics [I], narcotics or seda- tives [N], chemotherapeutic agents [C], and heparin or anticoagu- lants [H]. The Clinical Excellence Commission also lists an ‘other’ category for high-risk medications that may be identified at the local health district or facility level. Our local health district has further considered paracetamol and combinations, and antidepres- sants or antipsychotics to be high-risk.

Outcomes

The primary outcome measure was the proportion of patients with discrepancies on the self-administered medication history. The secondary outcome measure was the proportion of patients with a high-risk medication discrepancy based on the Clinical Excellence Commission list of medications.

Data analysis

All data were reported descriptively. Categorical variables, including primary and secondary outcomes were reported as pro- portions. Continuous and ordinal variables were reported as means with standard deviations or medians with interquartile ranges as appropriate. A multivariable Poisson regression analysis was conducted to determine the association between the predictor variables (number of medications, sex, age, completion of community pharmacy information field, completion of allergy field, completion of general practitioner information field) and the dependent variable of number of discrepancies. Univariate analyses were first conducted. Variables with a p < 0.2 were entered into the multivariable model. The association was reported as incident rate ratios (IRR) with 95% confidence intervals. The association between number of medications and number of dis- crepancies were depicted via a scatterplot. All analyses were con- ducted using Stata 15 (Stata Corporation, College Station, Texas).

1. Results
   1. Characteristics of the study cohort

A total of 200 patients were approached for consent. Of these, 186 consented for participation in the study and 176 completed the MYMEDS form. Subsequently, 10 were excluded because they did not meet inclusion criteria. The form was checked for accuracy via a BPMH for 138 patients, which was the final study cohort. The others (n = 28) could not be checked due to availability of the phar- macists, patient transfer to the ward before the pharmacist could conduct a BPMH in the ED, or subsequent discharge. The mean age was 62 +- 19 years and 52% (n = 72/138) were male. The median number of medications listed on the MYMEDS form by patients was 3 (IQR 1 to 6). However, the number of medications after the BPMH was taken was 6 (IQR 3 to 8) (p < 0.001). Most patients also completed the allergy section (88%, n = 121/138), general practitioner name or contact details (95%, n = 131/138, community pharmacy name or contact details (77%, n = 106/138).

Main results

In the final cohort, 37% (n = 51/138) had an accurate medication list (i.e. drug names), and 25% (n = 34/138) had an accurate medi- cation list and regimen (Fig. 1). Overall, the total number of discrepancies was as follows: 0 (25%, n = 34/138), 1 (34%, n = 47/138), 2 (11%, n = 15/138), and >=3 (30%, n = 42/138). The

number of patients with 0, 1, 2, or >=3 medication list discrepancies (i.e. wrong or omitted drug) were 37% (n = 51), 28% (n = 38), 9% (n = 13), and 26% (n = 36), respectively. Of the patients with correct medication lists (i.e. drug names), the number of patients with 0, 1, 2, or >=3 regimen discrepancies were 69% (n = 35/51), 20% (n = 10/51), 4% (n = 2/51), and 7% (n = 4/51), respectively. In the multivariable Poisson regression analysis, the number of medica-

Fig. 2. Scatterplot of number of medications versus number of discrepancies.

Table 1

Clinical excellence commission high-risk medication discrepancies.

Clinical Excellence Commission Medication class Number

A Anti-infectives 6

P Potassium and electrolytes 0

I Insulin or antidiabetics 10

N Narcotics or sedatives 7

C Chemotherapeutic agents 1

H Heparin or anticoagulants 7

Local health district modifications

Other Paracetamol and combinations 10

Other Antidepressants or 7

antipsychotics

tions (IRR 1.11, 95% CI 1.09 to 1.14, p < 0.001), female (IRR 1.51, 95% CI 1.18 to 1.92, p = 0.001), and missing community pharmacy

information (IRR 2.10, 95% CI 1.64 to 2.68, p < 0.001) were signifi- cantly associated with increased rate of discrepancies (Fig. 2). Overall, 20% (n = 28) of patients had one or more high-risk medica- tion discrepancies (A PINCH) (Table 1). When incorporating the local modifications to the pneumonic, 26% (n = 36) had one or more high-risk medication discrepancies.

1. Discussion

Fig. 1. Flow diagram for selection of study cohort.

The most important finding in this study was that only 25% of patients had an accurate MYMEDS form in the ED and 20% had dis- crepancies involving a high-risk medication. The results highlight that while a self-administered form may be feasible in this setting, the accuracy must be verified via a BPMH to mitigate the risk of Patient harm. Previous studies have shown that the medication history process in the ED is error prone, and even lists obtained by ED professionals may be inaccurate when compared to a BPMH by a pharmacist [11]. There may be also discordance between patient reported medications and those from pharmacy dispensing records [12]. Our results provide additional details regarding the type of discrepancies (i.e. medication list versus regimen), and fac- tors that could help triage patients who are at higher risk for errors. The use of a self-administered medication history form is very common in general practitioners‘ offices and other ambulatory care settings. However, this practice has not been widely imple- mented in the ED. This is primarily due to logistical considerations in the ED where patients may be more acutely ill, have mental

status changes, or lack manual dexterity. The ED visit may occur suddenly without prior planning for patients to gather their med- ications or reflect upon the medications they are taking. Also, patients may not consider the importance of some medications during their acute illness.

In a cross-sectional study in an urban ED in the United States, 70% (n = 249/354) of patients who were approached by Research staff completed a self-administered medication history form [5]. This increased to 76% after excluding those who refused to partic- ipate. The authors suggested that a self-administered medication history form was feasible and could reduce redundancies in the history-taking process. However, the form used in this study only elicited the listing of medication names from patients. The route, dose or frequency was not required. This could affect the utiliza- tion of the form. Also, the study did not measure the accuracy of the medication lists. In contrast to this, our form included both the medication names and regimens, which may be more difficult for patients. We also checked the accuracy of the lists by verifying it using a BPMH by pharmacists. Although 37% of patients in our study listed accurate medication names, only 25% listed accurate names and regimens.

Similar to our investigation, a study conducted in an ED in Canada has evaluated the accuracy of self-administered medica- tion history forms [6]. Discrepancies between the self- administered form and the BPMH were classified based on severity (none, minor, moderate or severe). Of the 160 patients in the study, 19% had none, 78% had minor, 15% had moderate, and 3% had sev- ere discrepancies. The authors concluded that the majority of dis- crepancies have limited clinical significance. In our study, we did not categorize discrepancies in this manner given the inherent subjectivity involved with such as assessment. Instead we simply considered which medications were high-risk according to the Clinical Excellence Commission. Using this definition, we found that 20% of patients had at least one high-risk discrepancy, which is concerning. The aforementioned Canadian study also only included patients who were discharged directly from the ED (i.e. non-admitted patients). However, in our ED and many others, a medication history is prioritized for those patients who are admit- ted, which was the population we investigated. Our sample may reflect those patients with a relatively higher acuity compared to the study mentioned above.

Previous studies have shown that the number of medications may be associated with Medication errors or discrepancies [6,13]. Thus, the finding of the significant association between number of medications and number of discrepancies is not surprising. However, according to our multivariable analysis, age was not sig- nificantly associated with discrepancies. Although we expect the number of medications to increase with age, the results high- lighted that the number of medications taken is a better predictor than age. Based on our clinical experience we had considered that patients who do not provide their pharmacy information may have less of a relationship with a particular community pharmacy and may be less thorough with their medication history. We showed that missing community pharmacy details also increased the rate of discrepancies. Although this finding should be replicated, this may be an important surrogate in the future for those at-risk for

medication errors. The significant association between sex and dis- crepancies was surprising (i.e. more discrepancies with females). We cannot identify any reasons of why this would occur, and it requires further investigation.

The study has a few important limitations. The selection of patients for whom a BPMH was taken was a convenience sample (i.e. when pharmacist was present in the ED) and is susceptible to selection bias. Although the pharmacist only completed a BPMH on admitted patients, it is possible that some patients were subse- quently directly discharged from the ED due to changes in decision-making with regard to patient disposition. Unfortunately, we did not keep track of these patients. The number of discrepan- cies is based on a comparison with a BPMH. However, a BPMH is not synonymous with a completely accurate medication history. This limitation is faced by most studies in this domain of research and alternatives to a BPMH do not exist in the real-world setting. The study was conducted in one ED in Australia and should be extrapolated with caution. We did not consider herbals or supple- ments for this study, which could potentially increase the rates of discrepancy. The results could vary by socioeconomic and health literacy levels, which were not measured in the study. The Clinical Excellence Commission pneumonic for high-risk medications was not specifically designed for the ED setting. Some medications may be considered high-risk only in certain circumstances, rather than for all patients. The number of high-risk medications can vary based on context of the patient presentation. However, we incor- porated this list because is commonly used across health care settings.

In conclusion, patient self-administered medication history forms in the ED have a high rate of error and a substantial propor- tion include high-risk medications. While a self-administered medication history forms may be a starting point for an initial medication list and regimen, the accuracy of the forms should be verified by a BPMH.

Conflicts of interest and source of funding

None.

Acknowledgements

None.

Abstract publication/presentation

None.

Author contribution statement

AW, MS, SA, and JE conceived of and designed the study. AW, MS, and SA collected the data. AP analyzed the data. AW and AP wrote the first draft of the manuscript and all authors contributed to it and approved its revision.

Appendix A. MYMEDS form

References

1. American Society of Health-System Pharmacists. ASHP statement on pharmacy services to the emergency department. Am J Health-Syst Pharm;65:2380-3.
2. Tully MP. Prescribing errors in hospital practice. Br J Clin Pharmacol 2012;74:668-75.
3. Meguerditchian AN, Krotneva S, Reidel K, Huang A, Tamblyn R. Medication reconciliation at admission and discharge: a time and motion study. BMC Health Serv Res 2013;13:485.
4. Hummel J, Evans PC, Lee H. Medication reconciliation in the emergency department: opportunities for workflow redesign. Qual Saf Health Care 2010;19:531-5.
5. Witting MD, Hayes BD, Schenkel SM, Drucker CB, DeWane MP, Lantry 3rd JH, et al. Emergency department medication History taking: current inefficiency and potential for a self-administered form. J Emerg Med 2013;45:105-10.
6. MacDonald N, Manuel L, Brennan H, Musgrave E, Wanbon R, Stoica G. Reliability of best possible medication histories completed by non-admitted patients in the emergency department. Can J Hosp Pharm 2017;70:263-9.
7. Continuity of Medication Management. Clinical excellence commission, http:// www.cec.health.nsw.gov.au/patient-safety-programs/medication- safety/continuity-of-medication-management.
8. Best possible medication history. Canadian Patient Safety Institute. Available at: https://www.patientsafetyinstitute.ca/en/Topic/Pages/Best-Possible- Medication-History.aspx (Accessed 13 February 2019).
9. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009;42:377-81.
10. Clinical Excellence Commission. High-risk medicines, http://www.cec.health. nsw.gov.au/patient-safety-programs/medication-safety/high-risk-medicines/ A-PINCH.
11. Baena Parejo MI, Juanes Borrego AM, Altimiras Ruiz J, Crespi Monjo M, Garcia- Pelaez M, Calderon Hernanz B, et al. Medication list assessment in Spanish hospital emergency departments. J Emerg Med 2015;48:416-23.
12. Meyer C, Stern M, Woolley W, Jeanmonod R, Jeanmonod D. How reliable are patient-completed Medication reconciliation forms compared with pharmacy lists? Am J Emerg Med 2012;30:1048-54.
13. Patanwala AE, Warholak TL, Sanders AB, Erstad BL. A prospective observational study of medication errors in a tertiary care emergency department. Ann Emerg Med 2010;55:522-6.